Trust and the Choice of Knowledge Transfer Mechanisms in Clusters
This study examines the impact of trust on the use of knowledge transfer mechanisms of cluster firms by deriving hypotheses from a relational governance perspective. Specifically, we analyze the influence of trust on the use of high information-rich (HIR) knowledge transfer mechanisms in cluster relationships. Based on the relational view of governance, trust may influence the choice of knowledge transfer mechanisms of the cluster partners in the following way: If trust reduces relational risk, more trust reduces the firms‘ use of HIR-knowledge
transfer mechanisms. If trust increases knowledge sharing between the cluster partners, it increases the firms‘ use of HIR-knowledge transfer mechanisms. In addition, we hypothesize
that interorganizational experience moderates the relationship between trust and the use of knowledge transfer mechanisms. We test the hypotheses by using data from the Italian textile and fashion sector.
Knowledge transfer mechanisms, trust, relational view of governance, cluster relationships
1 Marijana Srećković, Department of Real Estate Development, Faculty of Architecture and Planning, Vienna University of Technology, Gusshausstrasse 30, A-1040 Vienna, Austria; Email: firstname.lastname@example.org 2 Josef Windsperger, Center for Business Studies, University of Vienna, Brünner Str. 72, A-1210 Vienna, Austria; Email: email@example.com
Knowledge transfer between cluster partners is a key to gaining and sustaining competitive advantage (e.g. Maskell and Malmberg 1999; Driffield and Munday 2000; Levin et al. 2002; Hult et al. 2004; Li 2004). The success of cluster relationships depends on the effectiveness of the transfer of the know-how between cluster partners. Thereby plays trust a critical role for the performance of affiliated firms (Liao 2010). This study examines the impact of trust on the choice of knowledge transfer mechanisms of cluster firms by developing hypotheses based on the information richness theory and the relational view of governance.
Information richness theory offers information richness as a criterion to evaluate the transfer capacity of the communication media as knowledge transfer mechanisms (Daft and Lengel 1986; Büchel and Raub 2001; Sexton et al. 2003; Sheer and Chen 2004; Vickery et al. 2004). Information richness increases with the following attributes of a knowledge transfer mechanism: feedback capability, availability of multiple cues (voice, body, gestures, words), language variety, and personal focus (emotions, feelings). In cluster relationships, knowledge transfer mechanisms with a relatively higher degree of information richness include seminars, workshops, committees, conference meetings, and visits. Knowledge transfer mechanisms with a relatively lower degree of information richness include written documents, fax, email, intra- and internet and other electronic media.
According to the relational view of governance (e.g. Gulati 1995; Dyer and Singh 1998; Zaheer et al 1998; Poppo and Zenger 2002; Gulati and Nickerson 2008), trust reduces relational risk and increases information sharing and therefore influences the use of knowledge transfer mechanisms. We hypothesize two trust effects (Gorovaia and Windsperger
2011): If trust reduces relational risk, it will reduce the cluster partners‘ need to use more
HIR-knowledge transfer mechanisms. On the other hand, if trust increases knowledge sharing, it will increase the cluster partners‘ use of HIR-knowledge transfer mechanisms.
Although many researchers have examined the problem of knowledge transfer in network relationships in the last two decades (e.g. Nonaka 1994; Simonin 1999a, 1999b; Albino et al. 1999; Bresman et al. 1999; Argote and Ingram 2000; Altinay and Wang 2006; Jensen and Szulanski 2007; Szulanski and Jensen 2006; Haas and Hansen 2007; Baccera et al. 2008; van Wijk et al. 2008; Minguela–Rata et al. 2010; Winter et al. 2011), this literature does not investigate the determinants of the choice of knowledge transfer mechanisms in interorganizational networks. To the best of our knowledge, the works of Inkpen and Dinur
), as well as Srećković (1998), Murray and Peyrefitte 2007, Windsperger and Gorovaia (2011
and Windsperger (2011) are exemptions. They develop and test a knowledge-based view by analyzing the relationship between knowledge characteristics and knowledge transfer mechanisms used in joint ventures, franchising and cluster relationships. According to the knowledge-based theory, the tacitness of partner knowledge determines the degree of information richness of the knowledge transfer mechanisms. In this study, we extend the knowledge-based view of the choice of knowledge transfer mechanisms of Srećković and
Windsperger (2011) by considering trust as an additional explanatory variable of the cluster firm‘s knowledge transfer strategy. Our empirical study tests hypotheses by utilizing primary data from the Italian textile and fashion cluster that enables us to estimate the influence of trust on the knowledge transfer strategy of the cluster firms.
The article is organized as follows: Section two gives an overview of the relevant literature. Section three derives the hypotheses form the relational view of governance.
Section four tests the hypotheses with data from the Italian cluster. Section five discusses the results and derives some conclusions.
2. Trust and Choice of Knowledge Transfer Mechanism in Clusters
If the know-how of the cluster firms is codifiable, trust has no or only weak influence on the impact of knowledge attributes on the use of knowledge transfer mechanisms, because exchange hazards are very low and the cluster firm can explicitly specify the relevant knowledge in the contract (Levin and Cross 2004; Gulati and Nickerson 2008). If the know-how of the cluster firms is tacit, the contracts between the cluster partners are very incomplete and the cluster firms have difficulties to successfully apply the partner-specific knowledge. Consequently, under highly tacit knowledge, trust has a major impact on knowledge transfer (Levin et al. 2002). Based on the relational view of governance (e.g. Macneil 1981; Zaheer and Venkatraman 1995; Dyer and Singh 1998; Lazzarini et al. 2008; Gorovaia and Windsperger 2011), we can differentiate two perspectives regarding the impact of trust on the use of knowledge transfer mechanisms:
Reduction of relational risk: Trust reduces the knowledge transfer hazards by
decreasing relational risk (Gulati 1995; Yu et al. 2006). When the cluster partners trust each other, their tolerance level of perceived risk will be higher, and the cluster firms will more probably select knowledge transfer mechanisms with a lower degree of information richness (Lo and Lie 2008). Hence, under high trust, the cluster firms are likely to use less HIR-knowledge transfer mechanisms because in this low relational risk situation low information-rich knowledge transfer mechanisms facilitate sufficient knowledge sharing. Conversely, when distrust exists between the cluster partners, their tolerance level of perceived risk will be
lower, and the cluster firms will be more likely to select knowledge transfer mechanisms with a higher degree of information richness that transfer more knowledge in order to reduce the degree of relational uncertainty. We derive the following hypothesis:
H1: The higher trust, the less likely is the use of HIR-knowledge transfer mechanisms.
Increase of Knowledge sharing: Trust overcomes communication barriers and
facilitates knowledge sharing and increases therefore the use of all modes of knowledge transfer (Blomqvist et al. 2005; Yeh et al. 2006; Seppänen et al. 2007; Bohnet and Baytelman 2007; Lazzarini et al. 2008). In addition, more communication due to the use of more HIR- knowledge sharing mechanisms may lead to more trust between the cluster partners (Anderson and Narus 1990; Dyer and Chu 2000; Blomqvist et al. 2005; Fink and Kraus 2007; Ben-Ner and Puttermann 2009). Consequently, under high trust, the cluster firms use more HIR- knowledge transfer mechanisms because trust creates an incentive for intense and open communication. As a result, we can derive the following hypotheses:
H2: The higher trust, the more likely is the use of HIR-knowledge transfer
Interorganizational Experience as Moderator
According to Reagans and McEvily (2003), the frequency of communication through interorganizational experience influences the knowledge transfer process. We hypothesize that interorganizational experience moderates the relationship between trust and the use of
knowledge transfer mechanisms. We can differentiate two effects: (A) If trust reduces relational risk, more experience with the network partner results in a stronger decrease in the use of HIR-knowledge transfer mechanisms when interorganizational experience increases. (B) If trust overcomes communication barriers and facilitates knowledge sharing, interorganizational experience increases the positive impact of trust on the use of HIR-knowledge transfer mechanisms. Therefore, depending on the role of trust as a relational risk reduction or a knowledge sharing mechanism, we can derive the following hypotheses:
H1A: The negative impact of trust on the use of HIR-knowledge transfer mechanisms
increases with interorganizational experience.
H2A: The positive impact of trust on the use of HIR-knowledge transfer mechanisms
increases with interorganizational experience.
3. Empirical Analysis
3.1. Sample and Data Collection
The empirical study uses data from the Italian textile and fashion industry. Italian industrial districts are a very important contributor to the Italian Economy, and considering
3the fashion and textile industry, Italy is one of the leading exporting countries in this field. In
2011, textile and fashion districts have accounted for 28.8% of the working population in Italy,
4employing about 537.435 people.
3 see http://mefite.ice.it/settori/Tessile.aspx?idSettore=02000000 [retrieved 20.11.2011] 4 see http://www.istat.it/en/ [retrieved 20.11.2011]
The empirical setting for testing these hypotheses is the Italian textile and fashion cluster situated in the Province of Prato in Tuscany. In 2009, the textile and clothing sector in the Prato district had an estimated workforce of 30.200 people and 7.582 business firms which accounted for a turnover of 3,872 Million Euros in that sector. ―Prato is one of the areas in
Central and Northeast Italy (the so-called ?Third Italy`) where centuries-old craft skills have successfully merged with modern industrial growth. Originating between the 19th and 20th centuries, the industrialization process underwent a rapid acceleration after World War II and was definitely established by the 1970s. During this period of development, Prato grew to become Europe‘s most important textiles and fashion centre, and the most advanced example — or prototype — of that particular form of organization of production that is the industrial district. One feature of industrial districts, and of the Prato district as well, is the specialization and distribution of work among small business firms; this segmentation finds its recomposition in a ?culturally and socially constituted` local market whose competitiveness is based more on the economical aspects of the area itself than on those of the single
We started our empirical work by analyzing textile and fashion companies working in Italian industrial districts. First, we contacted exclusively companies from the fashion cluster situated in the Prato district. The identification of cluster firms was based on two sources: (1)
6the online data bases (e.g., ―Unione Industriale Pratese‖) and (2) the Italian Chamber of
Commerce. In total, 426 residential cluster firms were contacted by mail. 144 companies
5 see http://www.ui.prato.it/unionedigitale/v2/english/presentazione%20distretto%20inglese.pdf [retrieved
20.11.2011] 6 see http://www.ui.prato.it/unionedigitale/v2/default.asp [retrieved 20.11.2011]
accessed the online questionnaire, but only 34 firms responded to most of the questions. Despite several attempts, ranging from multiple reminders to non-respondents and personal contacts via telephone, the response rate remained low. In order to increase the response rate and enlarge the sample, it was necessary to contact firms from other clusters as well. For this purpose, the so-called ―snowball technique‖ (Churchill and Iacobucci, 2005) was used. A
leading multinational fashion corporate group which is in cooperation with retailers and producers in the Italian industrial districts was contacted. General managers of the single affiliates were asked to contact exclusively with executive directors of target cluster firms, and to spread the questionnaire among cluster partners who might be interested in cooperating. General managers and executive directors were judged to be the most suitable respondents, or key informants, as they are the top decision makers in the company regarding the organization of the knowledge transfer between the partner firms. Key informants should occupy roles that make them knowledgeable about the issues being researched (John and Reve 1982). This procedure led to additional 131 questionnaires, i.e., questionnaires in which the majority of questions apart from the general company description have been answered. Unfortunately, the online questionnaire tool allowed skipping single questions or question batteries, thus the problem occurred that some respondents answered the questionnaire only in parts. However, the extension of the sample led to a satisfying sample size for all analyses. The questionnaire took approximately 10 minutes to complete on the average. We received 118 completed responses - a response rate of 27.70 %. We examined the non-response bias by investigating whether the results obtained from the analysis were driven by differences between the group of respondents and the group of non-respondents. Non-response bias was estimated by comparing early versus late respondents (Armstrong and Overton 1977), where late
respondents serve as proxies for non-respondents. No significant differences emerged between the two groups of respondents. In addition, based on Podsakoff et al. (2003), we used Harman‘s single-factor test to examine whether a significant amount of common method variance exists in the data. After we conducted factor analysis on all items and extracted more than one factor with eigenvalues greater than one, we felt confident that common method variance is not a serious problem in our study.
To test the hypotheses, the following variables are important: knowledge transfer mechanisms, trust, and control variables (see Appendix).
3.2.1 Knowledge Transfer Mechanisms
Our study conceptualizes information richness of knowledge transfer mechanisms in accordance with Daft and Lengel‘s approach (Daft and Lengel 1984). We measure high
information richness by the extent to which the partner firms use face-to-face knowledge transfer mechanisms, such as committees and formal meetings. The general managers were asked to rate the use of these knowledge transfer mechanisms on a five-point scale. The higher the score, the higher is the company's use of these HIR-knowledge transfer mechanisms (HIR) (see Appendix).
Trust (TRUST): According to the relational view of governance, trust may influence the use of knowledge transfer mechanisms in two ways: Under the substitutability view, trust is a substitute for the use of formal knowledge transfer mechanisms (Gulati 1995; Yu et al. 2006). It mitigates the knowledge transfer hazards and hence reduces the extent of formal knowledge transfer mechanisms (Lo and Lie 2008). Consequently, cluster companies are likely to use less HIR-knowledge transfer mechanisms when trust exists between the cluster partners, and use more HIR when mistrust exists. Under the complementarity view, trust facilitates knowledge sharing and increases the use of all knowledge transfer modes (Seppänen et al. 2007; Liao 2010). Therefore, under a high level of trust, cluster partners use more HIR-knowledge transfer mechanisms because trust creates an incentive for intense communication. TRUST was measured with a five-items scale (see Appendix) (Cronbach alpha = 0.89).
3.2.3. Control Variables
Complexity (COMPLEX): Kogut and Zander (1993, 633) define complexity ―as the
number of critical and interacting elements embraced by an entity or activity‖. Similarly,
Sorenson et al. (2006) define complexity in terms of the level of interdependence inherent in the subcomponents of a piece of knowledge (see Simonin 1999a,b). When the system knowledge is more complex, it is considered more tacit. Applied to the cluster relationships, complexity is high when the application of the partner knowledge requires a large number of